The rise of “artificial historians”: AI as a human record manager

Artificial intelligence is becoming a historian of humanity in documenting and recording collective social data on unprecedented scale.

However, the careless role of AI as a memory keeper raises deep concerns for historians today. Unlike human historians who explicitly document methodologies, AI systems create historical archives of the future without any significant transparency regarding source selection, weighting, and interpretation methods.

This undermines the fundamental principles of historical scholarships, and its methodology should be visible and contested. In his new book, “Artificial Historian,” historian Marnie Hughes Warrington explores how AI systems change historical records.

The author argues that AI is already deeply involved in creating history, producing “most of the history created all over the world.” Rather than viewing this as nothing more than a threat, the author encourages historians to see these systems as an opportunity to engage in AI development to reflect the historical complexity.

History and subtle understanding

Hughes-Warrington has submitted concerns about data collection, particularly biases in “uneven and unfair collection of information about the past.” When AI systems train these biased historical records, they risk amplifying and perpetuating historical inequality, potentially solidifying problematic narratives for future generations.

In addition to this, some historical information may be computable or unread by AI tools, providing incomplete images.

Similar to concerns about information gathering and transparency, Hughes Warrington points out that AI misses the nuances of historical storytelling that humans essentially embrace.

She means that historical claims made by scholars and historians are by no means completely or completely true, but are “partially grounded.” This complex understanding of historical information, or historical “truth,” presents the challenges of AI systems trained to provide a definitive answer.

When asked about the topic of world history, AI platforms tend to give similar traditional responses that show a limited view of history, indicating that AI systems lack a nuanced understanding of the historical context that human historians develop through years of research, suggests Hughes Warrington.

“Information from the past may not be available or even computationally available, may not be presented in a way that makes use of or combination of data sets difficult or impossible,” explains Hughes Warlington.

“The context of data collection can also be ignored. If you know that information has been collected about people in financial or judicial distress, would you use it without thinking about their experience?

ai stays here

“This cry from history and absorption into future, fiction, or geopolitics means that the historical expertise needed to make AI more effective and fair is overlooked. AI is not a threat to history when you look at the invitations involved in its creation,” she explains.

“By bringing historical expertise to AI development, we can create more effective and fair artificial historians while maintaining the critical thinking and contextual understanding that defines quality historical scholarships.”

The text ultimately suggests that making history is a complex, interpretive process that cannot be reduced to simple algorithms or rules. Hughes-Warrington challenges historians and AI engineers to think more deeply about how history is defined and created.

“If history is a problem, history is also a solution,” concludes Hughes Warrington.